@context dict | @type string | conformsTo list | name string | description string | alternateName string | url string | version string | datePublished timestamp[s] | keywords list | license string | isLiveDataset bool | creator list | publisher dict | citeAs string | citation string | sameAs list | isBasedOn list | distribution list | rai:dataCollection string | rai:dataUseCases string | rai:dataLimitations string | rai:dataBiases list | rai:personalSensitiveInformation string | rai:dataSocialImpact string | rai:hasSyntheticData string |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
{
"@language": "en",
"@vocab": "https://schema.org/",
"citeAs": "cr:citeAs",
"column": "cr:column",
"conformsTo": "dct:conformsTo",
"cr": "http://mlcommons.org/croissant/",
"rai": "http://mlcommons.org/croissant/RAI/",
"data": {
"@id": "cr:data",
"@type": "@json"
},
"dataType": {
"@id": ... | sc:Dataset | [
"http://mlcommons.org/croissant/1.0",
"http://mlcommons.org/croissant/RAI/1.0"
] | quantum_ground_state_ood_benchmark | Beyond In-Distribution Generalization: A Benchmark for Machine Learning on Ground-State Property Prediction of Quantum Systems. The benchmark contains five families of quantum many-body Hamiltonians: 1D Transverse-Field Ising (tfim_1d), 1D XXZ (xxz_1d), 1D Heisenberg with random local Z fields (heisenberg_local), 1D He... | Quantum-OOD-GS | https://huggingface.co/datasets/hdkqzpmta/quantum-ood-benchmark | 1.0.0 | 2026-05-07T00:00:00 | [
"quantum many-body physics",
"ground state",
"machine learning",
"out-of-distribution generalization",
"benchmark",
"transverse-field Ising model",
"XXZ model",
"Heisenberg model",
"classical shadows"
] | https://creativecommons.org/licenses/by/4.0/ | false | [
{
"@type": "Person",
"name": "Anonymous Authors",
"affiliation": {
"@type": "Organization",
"name": "Anonymous Institution"
}
}
] | {
"@type": "Organization",
"name": "Anonymous Institution"
} | @inproceedings{Anonymous_quantum_ood_2026, title={Beyond In-Distribution Generalization: Benchmarking Machine Learning for Ground State Property Prediction of Quantum Systems}, author={Anonymous Authors}, booktitle={NeurIPS Datasets and Benchmarks Track}, year={2026} } | Anonymous Authors (2026). Beyond In-Distribution Generalization: Benchmarking Machine Learning for Ground State Property Prediction of Quantum Systems. NeurIPS Datasets and Benchmarks Track. | [
"https://github.com/hsinyuan-huang/provable-ml-quantum"
] | [
{
"@type": "sc:Dataset",
"name": "provable-ml-quantum",
"url": "https://github.com/hsinyuan-huang/provable-ml-quantum",
"description": "Source of the heisenberg_2d family of this benchmark. We re-host the L x 5 (L in {4,...,9}) 2D Heisenberg classical-shadow data and ground-state observables release... | [
{
"@type": "cr:FileObject",
"@id": "heisenberg_2d_N20_s100_m1000.npz",
"name": "heisenberg_2d_N20_s100_m1000.npz",
"description": "Ground-state data for family 'heisenberg_2d' (family code 4) with system size N=20. Contains 100 Hamiltonian instances; per-instance classical-shadow snapshots are store... | The benchmark mixes data of two provenances: (1) The four 1D families (tfim_1d, xxz_1d, heisenberg_local, heisenberg_longrange) at every system size N in {20,40,60,80,100} were generated independently by the benchmark authors using ITensors/ITensorMPS DMRG, with classical-shadow snapshots produced by random single-qubi... | Intended for benchmarking ML models that predict properties of quantum ground states - in particular for measuring out-of-distribution generalisation across system size, Hamiltonian family and physical regime. The dataset deliberately does not ship a fixed train/test split; users must declare their OOD protocol when re... | All systems are spin-1/2 with at most N=100 sites; results may not extrapolate to fermionic systems or to the thermodynamic limit. Classical-shadow shot counts are fixed (M=1000 or 1024) and finite-shot noise is part of the data. | [
"Sampling distributions over Hamiltonian control parameters were chosen by the benchmark authors and reflect specific design choices: tfim_1d uses h ~ U[0.5, 2.0] (skewing toward the paramagnetic side of the critical point at h=1.0); xxz_1d uses Delta ~ U[-2.0, 2.0] (oversampling the gapless XY phase relative to a ... | None. This dataset contains no personal, demographic, biometric, health, behavioral, or otherwise human-subject data. All entries are numerical results of computer simulations of quantum many-body Hamiltonians (DMRG ground-state observables and synthetic classical-shadow measurement outcomes). No personally identifiabl... | The dataset is intended for methodological machine-learning research on out-of-distribution generalization for ground-state property prediction of quantum many-body systems. We do not foresee direct social or societal harms from its release: it contains no human-subject data, encodes no demographic features, and the un... | Yes. The entire dataset is computer-generated. Hamiltonian instances are sampled from predefined random distributions (see rai:dataBiases for the per-family sampling protocol); ground-state observables are computed numerically with DMRG (ITensors/ITensorMPS for the four 1D families) or inherited from the open-source re... |
README.md exists but content is empty.
- Downloads last month
- 69